This repository is dedicated to the Capstone Project conducted at Paris School of Economics in colloboration with researchers from Université Gustav Eiffel (formerly known as IFSTTAR).
The project consists of data analysis of public transport data collected on Moovizy2 application launched by Transdev.
The project is under a Non Discolsure Agreement. Therefore the dataset used for analysis has to remain undisclosed.
Client wants to know if people use Moovizy2 application as another tool of navigation or is it actually serving as a MaaS application.
If it turns out as the second case, client wants to know what is the prediction power of consumer behaviour on the total ticket sales volume.
Here consumer behaviour is the set of all the explanatory variables explained below.
Although I cannot provide the dataset, I can explain what kind of information does the dataset contain.
All the following variables below are aggregated in terms of date and hourly interval:
-
Total ticket sales volume (regressand)
-
Total ticket sales volume by regular users/irregular users
-
Total number of requests
-
Total number of requests made within 2-hour-lag (lag: gap between request time and departure time)
-
Total number of requests made within single-day-lag
-
Total number of requests made more-than-a-day-lag
-
Weekend indicator (1 if weekend 0 ifnot)
-
Average number of transfers of the fastest trips
-
Average speed of the fastest trips
Plot images are in the Plots folder.
Regression results won't be available due to the conditions of NDA.
Visualisation https://www.kaggle.com/subinium/dark-mode-visualization-apple-version